Plain Language Summary
Heavy precipitation events are large storms that can recharge freshwater reservoirs, but can also lead to hazardous outcomes such as flash floods. Therefore, understanding the impacts of climate change on such storms is critical. Here, a weather model similar to those used in weather forecasts is used to simulate heavy precipitation events in the eastern Mediterranean. A large collection of storms is simulated in pairs: (1) historic storms, known for their high impact, and (2) placing the same storms in a global warming scenario projected for the end of the 21st century. Using these simulations we ask how present-day storms would look like were they to occur at the warmer end of the 21st century. The future storms are found to produce much less rainfall compared to the historic ones. This decrease in rainfall is attributed mainly to the reduction in the area covered by storms’ rainfall, and happens despite increasing rainfall intensities. These results suggest that the region will be drier in the future with larger dry areas during storms; however, over short durations, it would rain more intensely over contracted areas – increasing local hazards associated with heavy precipitation events.
1 Introduction
Expected impacts of climate change on rainfall during heavy precipitation events (HPEs) have the potential to significantly alter their influence on future societies. Where precipitation variability is high, such as in Mediterranean and arid climates, the impact of individual HPEs in terms of both peril (e.g., Borga et al., 2014; Dayan et al., 2021; Raveh-Rubin & Wernli, 2016; Rinat et al., 2020; De Vries et al., 2013) and water resources (Flaounas et al., 2021; Nasta et al., 2018; Samuels et al., 2009; R. G. Taylor et al., 2013) is great, and reliable projections of HPEs are needed (e.g., Sillmann et al., 2021).
Individual HPEs are controlled by specific large scale and synoptic circulation patterns. However, projected changes in the atmospheric circulation are highly uncertain across global climate models (GCMs) due to the wide variety of factors at play (Shepherd, 2014). Furthermore, climate change impact on HPEs can be quite different from the well-studied impact on the mean rainfall or even on high precipitation percentiles (e.g., Donat et al., 2016; Kendon et al., 2018; Moustakis et al., 2021; O’Gorman, 2015; Pfahl et al., 2017; Trenberth et al., 2015).
Detailed projections of the regional rainfall during a specific event can only be provided by models that can explicitly resolve the convective processes governing precipitation during HPEs (e.g., Fosser et al., 2014). Indeed, convection-permitting models (CPMs) are more reliable than GCMs in simulating spatiotemporal precipitation patterns (Ban et al., 2014; Cannon & Innocenti, 2019; Crook et al., 2019; Kendon et al., 2014; Meredith et al., 2020; Poujol et al., 2020; Prein et al., 2015, 2017; Westra et al., 2014). Recent methodological and computing advances enable “climate” CPM simulations with long-term (~10 yr), large-scale (continental), and high resolution (a few kilometers) outputs with some groups already running ensemble simulations over specific regions (Chan et al., 2020; Coppola et al., 2020; Pichelli et al., 2021). These give probabilistic projections of changes in precipitation extremes with expectations to achieve better quantification of future HPEs (e.g., Kendon et al., 2014; Poujol et al., 2020). However, rare extreme or heavy precipitation events are, by definition, hard to assess even with such simulations (e.g., Fatichi et al., 2016; Kendon et al., 2021). Moreover, a few kilometers resolution may still not be sufficient to represent the local nature of convective clouds, especially when shallow convection is present (Kendon et al., 2021; Prein et al., 2015). Therefore, trying to provide reliable projections of the changes in rainfall patterns during HPEs will probably take many more years of improvement in climate modeling. A complementing approach, aimed at resolving extreme events and intra-event characteristics (Fowler, Ali, et al., 2021; O’Gorman, 2015), is to provide projections of specific high impact events either by identifying interesting events such as hurricanes over long-term simulations (Gutmann et al., 2018), or through the simulation of individual events known for their high-impact, such as snowstorms (G. Chen et al., 2020), tropical cyclones (J. Chen et al., 2020), or HPEs (Ferreira, 2021).
Pseudo global warming (PGW) is an emerging methodology for event-based projections, enabling assessment of the impacts of one or more meteorological parameters over local-scale weather events (Brogli et al., 2019; Fowler, Lenderink, et al., 2021; Moustakis et al., 2021; Prein et al., 2017; Sato et al., 2007; Schär et al., 1996). The PGW methodology imposes a certain climate change, e.g., temperature rise, over the initial and boundary conditions of a regional model, by prescribing the synoptic and larger-scale changes from GCMs, while allowing smaller scale features to develop freely within a downscaled modeled domain in a physically consistent manner. Further, projections of precipitation extremes under global warming scenarios commonly focus on daily resolutions (Donat et al., 2016; O’Gorman, 2015; Pfahl et al., 2017), which hinders the possible impact of short-duration extremes; only recently more studies have directed attention to changes expected over sub-daily or even sub-hourly extremes (Fowler, Ali, et al., 2021; Fowler, Wasko, et al., 2021; Morrison et al., 2019). However, to understand the potential effects of changes in precipitation extremes, not only their changing intensity and frequency are important, but also high-resolution changes in intra-event characteristics, such as the spatiotemporal organization of the storms (Li et al., 2018). This requires high-resolution analysis of many high-impact storms of different synoptic-scale circulations, as there is no guarantee that different HPEs behave the same way (Fowler, Ali, et al., 2021).
The goal of this study is to identify and quantify changes in rainfall patterns during HPEs induced by global warming, and to examine whether a common change emerges over a variety of HPEs. To do so, we exploit the case of the eastern Mediterranean (Sect. 2.1) to simulate a large number of HPEs using the PGW methodology with a very high spatiotemporal resolution, and explicitly consider space-time patterns of rainfall during the events over durations of 10-min to 24-h.
The paper is organized as follows: Section 2 describes the study region and outlines the modeling strategy and the analyses of rainfall patterns. We first demonstrate the expected changes for a specific HPE case (Sect. 3.1), and then examine changes in rainfall accumulation over a large set of HPEs (Sect. 3.2). Changes in specific rainfall properties are outlined in Sect. 3.2-3.4, with the unique role of the rain area shown in Sect. 3.3.1. Section 4 begins with a discussion of the event-based approach (Sect. 4.1) and continues with an examination of the change in rainfall patterns in future HPEs (Sect. 4.2). Our conclusions are presented in Sect. 5.
2 Study Region, Data, and Methods
2.1 Study Region
The focus here is on the eastern Mediterranean (Fig. 1), which (a) is expected to suffer from a large future decrease in total rainfall (Garfinkel et al., 2020; Giorgi & Lionello, 2008; Zappa et al., 2015), (b) may experience an increase in extreme precipitation occurrence (Alpert et al., 2002; Marra et al., 2021; Samuels et al., 2017), (c) is characterized by the least precipitation per capita in the world (Dirmeyer et al., 2009), and (d) is exposed to large rainfall variability (Morin, 2011). These characteristics result in a large dependency on HPEs, in terms of water resources and vulnerability to natural hazards; therefore, we explore here possible future changes in HPEs in the region, and disassemble them to their distinct hydrometeorological constituents. It is important to note there is currently no CPM with future projections available for the study region.
In the eastern Mediterranean (Fig. 1a), the Mediterranean climate abuts the semiarid to hyperarid climates characterizing the region to the south and east of the Mediterranean Sea. Yearly rainfall amounts drop from >1000 mm in the northern mountains, to <<100 mm at the southeast regions (Fig 1b). Summers are dry, and the rainy season is October to May, with a few rare exceptions in September and June (Yair Goldreich, 2012; Kushnir et al., 2017). The core of the rainy season is December-February (>65% of precipitation). However, the rainy season’s midpoint changes from the beginning of January near the Mediterranean Sea to the end of January farther inland (Y. Goldreich, 1994; Yair Goldreich, 1995). This reflects the important contribution of the warm Mediterranean Sea water to building up of Mediterranean Cyclones (MCs), the favorable synoptic condition prevailing during rainy days, generating >90% of all rainfall in the northern, wetter part of the region (Alpert & Shay-EL, 1994; El‐Fandy, 1946; Ziv et al., 2006). Other synoptic systems contribute relatively large rain amounts to the interior-desert area mainly during the transitional seasons (Armon et al., 2019; Dayan & Morin, 2006; Kahana et al., 2002), including the more frequent (a) active Red Sea troughs (ARSTs) (Ashbel, 1938; De Vries et al., 2013), occurring mainly in fall, and (b) less frequent disturbances in the Subtropical Jet sometimes termed Tropical Plumes or Active Subtropical Jet (Armon et al., 2018; Dayan & Abramski, 1983; Rubin et al., 2007; Tubi et al., 2017).